Merge pull request #14 from pratapvardhan/hist_func

ENH: Add basic support for pd.DataFrame.hist pd.Series.hist

Author: jakevdp

altair_pandas/__init__.py

@@ -1,5 +1,5 @@
 """Altair plotting extension for pandas."""
 __version__ = "0.1.0dev0"
-__all__ = ["plot"]
+__all__ = ["plot", "hist_frame", "hist_series"]
 
-from ._core import plot
+from ._core import plot, hist_frame, hist_series

altair_pandas/_core.py

@@ -87,6 +87,9 @@ def hist(self, bins=None, **kwargs):
             )
         )
 
+    def hist_series(self, **kwargs):
+        return self.hist(**kwargs)
+
     def box(self, **kwargs):
         data = self._preprocess_data(with_index=False)
         return (
@@ -190,6 +193,19 @@ def hist(self, bins=None, stacked=None, **kwargs):
             )
         )
 
+    def hist_frame(self, grid_columns=2, **kwargs):
+        data = self._preprocess_data(with_index=False)
+        data = data._get_numeric_data()
+        return (
+            alt.Chart(data)
+            .mark_bar()
+            .encode(
+                x=alt.X(alt.repeat("repeat"), type="quantitative", bin=True),
+                y=alt.Y("count()", title="Frequency"),
+            )
+            .repeat(repeat=list(data.columns), columns=grid_columns)
+        )
+
     def box(self, **kwargs):
         data = self._preprocess_data(with_index=False)
         return (
@@ -210,3 +226,11 @@ def plot(data, kind="line", **kwargs):
         raise NotImplementedError(f"kind='{kind}' for data of type {type(data)}")
 
     return plotfunc(**kwargs)
+
+
+def hist_frame(data, **kwargs):
+    return _PandasPlotter.create(data).hist_frame(**kwargs)
+
+
+def hist_series(data, **kwargs):
+    return _PandasPlotter.create(data).hist_series(**kwargs)

altair_pandas/test_plotting.py

@@ -150,3 +150,23 @@ def test_dataframe_boxplot(dataframe, with_plotting_backend):
     assert spec["encoding"]["x"]["field"] == "column"
     assert spec["encoding"]["y"]["field"] == "value"
     assert spec["transform"][0]["fold"] == ["x", "y"]
+
+
+def test_dataframe_hist_series(series, with_plotting_backend):
+    chart = series.hist()
+    spec = chart.to_dict()
+    assert spec["mark"] == "bar"
+    assert spec["encoding"]["x"]["field"] == "data_name"
+    assert "field" not in spec["encoding"]["y"]
+    assert spec["encoding"]["x"]["bin"] == {"maxbins": 10}
+
+
+def test_dataframe_hist_frame(dataframe, with_plotting_backend):
+    chart = dataframe.hist()
+    spec = chart.to_dict()
+    assert spec["repeat"] == ["x", "y"]
+    assert spec["columns"] == 2
+    assert spec["spec"]["mark"] == "bar"
+    assert spec["spec"]["encoding"]["x"]["field"] == {"repeat": "repeat"}
+    assert spec["spec"]["encoding"]["x"]["bin"] is True
+    assert "field" not in spec["spec"]["encoding"]["y"]